Proceedings of the 2017 7th International Conference on Education and Management (ICEM 2017)

Ensemble Radical Basis Function Neural Networks for Regression Based on Statistical Learning Theory

Authors
Liangzhi Gan, Dawei Jiang, Mi He
Corresponding Author
Liangzhi Gan
Available Online January 2018.
DOI
https://doi.org/10.2991/icem-17.2018.71How to use a DOI?
Keywords
Statistical learning theory; Radical basis function neural networks; VC dimension; Ensemble learning
Abstract
We proposed an algorithm to construct ensemble radical basis function neural networks for regression estimation. Taking full advantage of the characteristic of radial basis function, we calculated groups of approximate basis in Reproducing Kernel Hilbert Space (RKHS). The approximate basis could be used to represent all the samples by the way of linear combination. By this way, the weak learners of radial basis function neural network were built. But it was proved that the weak learners were not accurate enough. In order to get accurate and stable learning machine with better generalization ability, we proposed the Ensemble Radical Basis Function Neural Networks (ERBFNNs). Employing the sinc function, the proposed ERBFNNs have shown exciting outcomes as have come out at the end of the paper.
Open Access
This is an open access article distributed under the CC BY-NC license.

Download article (PDF)

Proceedings
Part of series
Advances in Economics, Business and Management Research
Publication Date
January 2018
ISBN
978-94-6252-463-7
ISSN
2352-5428
DOI
https://doi.org/10.2991/icem-17.2018.71How to use a DOI?
Open Access
This is an open access article distributed under the CC BY-NC license.

Cite this article

TY  - CONF
AU  - Liangzhi Gan
AU  - Dawei Jiang
AU  - Mi He
PY  - 2018/01
DA  - 2018/01
TI  - Ensemble Radical Basis Function Neural Networks for Regression Based on Statistical Learning Theory
PB  - Atlantis Press
SP  - 352
EP  - 355
SN  - 2352-5428
UR  - https://doi.org/10.2991/icem-17.2018.71
DO  - https://doi.org/10.2991/icem-17.2018.71
ID  - Gan2018/01
ER  -